NDMP restore issue

[Henning_General]

Hello People,

I want to restore old LTO3 tapes(written by an EMC Cellera) to same EMC Cellera

The ndmp host is configured. 

I have got a new Windows Server 2008R2 with NBU 7.5 installed. It´s my first time to restore with NDMP

In the Restore Manger I filled in the following configuration:

Server: My master server

Source Client: the ndmp filer 

Destination Client: the ndmp host, which is already configurated

Policy: NDMP

When I try to start the restore everything to differnt location, than this error will apear. Unable to connect to Netbackup host "myNdmphost"

Thanks for your help!

Henning

[Marianne]

Legacy log folders do not exist by default....

They need to be created.

Like you had to create bprd.
Only difference bpbrm does not need NBU to be restarted.

It seems ndmpagent log is pointing us in the right direction.

SERVER_PORT_WINDOW is trying non-reserved ports above 50000.

See ndmpagent log: 

.... -jobid 29 -nonrsvdports -S ge-2008 -browse_clnt defmoa07psge -ndmpport 53316

.... -jobid 30 -nonrsvdports -S ge-2008 -browse_clnt defmoa07psge -ndmpport 54511

Extract from NBU Ports Reference Guide:

NDMP server ports
The following list describes the port requirements to backup and restore an NDMP server:
■ TCP port 10000 must be open from the media server (DMA) to the NDMP filer
(tape or disk) for all types of NDMP operations; local, remote, and 3-way..
■ The NetBackup SERVER_PORT_WINDOW must be open inbound from the
filer to the media server for remote NDMP ..... 

See this TN to specify a range for SERVER_PORT_WINDOW and get your firewall admins to open this range.

SERVER_PORT_WINDOW option for NetBackup servers 
http://www.veritas.com/docs/000044042

[mnolan]

You should not use a single port for the SERVER_PORT_WINDOW as you will run out of ports AND opening the same server up on the same port can be a security threat.

We've clearly listed the SERVER_PORT_WINDOW as required for remote NDMP in the firewall guide and a technote stating where you can change SERVER_PORT_WINDOW along with its default range and usage.

NetBackup 6.x and 7.x firewall port requirements
Article:000008025 
Publish:September 14, 2015 
Article URL:http://www.veritas.com/docs/000008025

"To backup and restore NDMP filers:

Media server (DMA) to NDMP filer (tape or disk) requires TCP port 10000.

The SERVER_PORT_WINDOW is used inbound from the filer to the media server for remote NDMP and can also be used for efficient catalog file (TIR data) movement with local and 3-way NDMP. "

SERVER_PORT_WINDOW option for NetBackup servers
Article:000044042 
Publish:October 21, 2015 
Article URL:http://www.veritas.com/docs/000044042

Your ndmpagent log shows the error from the filer to connect back for data.

ndmp_data_connect failed, status = 23 (NDMP_CONNECT_ERR),28:NdmpDataSideNas::DataConnect,1

We have articles and technotes where backups/restores fail with this error and it is exactly as I have described it.

https://www-secure.symantec.com/connect/forums/ndmp-backup-failing

https://www.veritas.com/support/en_US/article.000021101

[Henning_General]

Hello Marianne,

is it possible to use only one Port for remote NDMP ?

Or I need a special (for example :default) port range ?

regards,

Henning

[Marianne · Accepted solution]

The NDMP server port is a single port as per extract from the NBU Ports Reference Guide.

As per mnolan's excellent post above - SERVER_PORT_WINDOW  cannot be a single port - you need a range.

Please see the TN in my post above. 
Show it to you Firewall Admins.
There is no other way.
Or move your master/media server to inside the firewall....

[mnolan]

Check the post regarding SERVER_PORT_WINDOW on where that setting is.

If you change it to a single port you will more than likely have issues (There is a reson it is a range.)

[Henning_General]

Hi Marianne,

thanks for your help.

I will move the ns480 (nas) into the backup network. After the implementation, I will inform you 

regards,

henning

[Henning_General]

Hi Marianne,

I moved the ns480 into the backup network. I´m able to connect to the NDMP Host.

What is the next step to restore the data on the ns480? 

Thanks,

Henning

[Marianne]

Does this mean the ns480 now have a new or addional IP address on the backup network?

Have you added this IP address to the hosts file on the master/media server using the same hostname as for the backups?
(Previous logs seem to indicate either defmoa07 or defmoa07psge.)

If you can connect / verify NBU comms using this hostname, you should be able to start the restore using the same hostname.

[Henning_General]

I updated the host file with the new IP. Im able to verify the ndmp host.  

How can I restore to the ns480 to a differnent folder location ? 

[Marianne]

I am not an expert on NDMP restores (I have very little 1st-hand experience).

I see the following in NBU NDMP Appliance doc:

For NAS filers that use NDMP V4, you can specify a different subfolder or file name.
For example, /vol/vol1/mydir/myfile can be restored to
/vol/vol2/mydir2/myfile_restored.

[Henning_General]

Have you got any how to document for me ?

Thanks for your help !

[jim_dalton]

You select the target folder as part of the restore setup, when youve pressed "Restore..." you get to choose.Jim

[Marianne]

There is no NBU documentation for redirected restore. Jim is the one with 1st-hand experience!

[Henning_General]

Hi Marianne ,

Now I get the following Error:

12/3/2015 8:56:26 AM - begin Restore
12/3/2015 8:56:26 AM - 1 images required
12/3/2015 8:56:26 AM - media 000399 required
12/3/2015 8:56:27 AM - restoring image defmoa_1230568092
12/3/2015 8:56:27 AM - Info bpbrm(pid=5664) defmoa is the host to restore to     
12/3/2015 8:56:27 AM - Info bpbrm(pid=5664) telling media manager to start restore on client    
12/3/2015 8:56:29 AM - Info bpbrm(pid=5536) defmoa is the host to restore to     
12/3/2015 8:56:29 AM - Info bpbrm(pid=5536) start tar32 on client        
12/3/2015 8:56:29 AM - Info tar32(pid=6992) Restore started.          
12/3/2015 8:56:29 AM - connected
12/3/2015 8:56:30 AM - Info bptm(pid=6968) Waiting for mount of media id 000399 (copy 1) on server ge-2008.
12/3/2015 8:56:30 AM - started process bptm (6968)
12/3/2015 8:56:30 AM - mounting 000399
12/3/2015 8:56:30 AM - Info bptm(pid=6968) INF - Waiting for mount of media id 000399 on server ge-2008 for reading.
12/3/2015 8:56:30 AM - requesting resource 000399
12/3/2015 8:56:30 AM - granted resource 000399
12/3/2015 8:56:30 AM - granted resource HP.ULTRIUM3-SCSI.002
12/3/2015 9:13:14 AM - Error bptm(pid=6968) error requesting media, TpErrno = Robot operation failed    
12/3/2015 9:13:54 AM - current media 000399 complete, requesting next resource HP.ULTRIUM3-SCSI.002:NetBackup:000399
12/3/2015 9:44:23 AM - Error ndmpagent(pid=6992) NDMP restore failed from path UNKNOWN      
12/3/2015 9:44:23 AM - Info ndmpagent(pid=6992) done. status: 150: termination requested by administrator     
12/3/2015 9:44:23 AM - Info bpbrm(pid=5664) got ERROR 98 from media manager      
12/3/2015 9:44:23 AM - Info bpbrm(pid=5664) terminating bpbrm child 5536 jobid=8       
12/3/2015 9:44:23 AM - Error nbjm(pid=3544) NBU status: 98, EMM status: The media server reported a system error
12/3/2015 9:44:24 AM - restored image defmoa_1230568092 - (error requesting media (tpreq)(98)); restore time 00:47:57
12/3/2015 9:44:24 AM - end Restore; elapsed time: 00:47:58
Restore error(2850)

[Marianne]

Waiting for mount of media id 000399 on server ge-2008 for reading.
granted resource HP.ULTRIUM3-SCSI.002
- Error bptm(pid=6968) error requesting media, TpErrno = Robot operation failed    
Error nbjm(pid=3544) NBU status: 98, EMM status: The media server reported a system error

Media server  ge-2008 cannot load tape 000399 into drive HP.ULTRIUM3-SCSI.002.

Actually - the robot reported a mount error...

Does the robot have a web interface?
Can you use this or robtest to test tape mount outside of NBU?

[Henning_General]

Hi,

the robot has got no web interface, but we can use robotest.

Which command/information you need ?

regards

[Marianne]

Simply use robtest to see if you can mount this tape into the drive.
Unmount it again and return to its slot.

You will need the slot number number for the restore tape and drive number (position in robot).

[Henning_General]

Im able to move the 3 tapes :

Initiating MOVE_MEDIUM from address 1002 to 502
move_medium failed
sense key = 0x5, asc = 0x3b, ascq = 0xe, MEDIUM SOURCE EL

Invalid command
m s1 d1
Initiating MOVE_MEDIUM from address 1000 to 500
MOVE_MEDIUM complete
s d
drive 1 (addr 500) access = 0 Contains Cartridge = yes
Source address = 1000 (slot 1)
Barcode = 000401
drive 2 (addr 501) access = 1 Contains Cartridge = no
drive 3 (addr 502) access = 0 Contains Cartridge = yes
Source address = 1002 (slot 3)
Barcode = 000399
drive 4 (addr 503) access = 1 Contains Cartridge = no
drive 5 (addr 504) access = 1 Contains Cartridge = no
drive 6 (addr 505) access = 1 Contains Cartridge = no
READ_ELEMENT_STATUS complete

Invalid command
m s2 d5
Initiating MOVE_MEDIUM from address 1001 to 504
MOVE_MEDIUM complete
s d
drive 1 (addr 500) access = 0 Contains Cartridge = yes
Source address = 1000 (slot 1)
Barcode = 000401
drive 2 (addr 501) access = 1 Contains Cartridge = no
drive 3 (addr 502) access = 0 Contains Cartridge = yes
Source address = 1002 (slot 3)
Barcode = 000399
drive 4 (addr 503) access = 1 Contains Cartridge = no
drive 5 (addr 504) access = 0 Contains Cartridge = yes
Source address = 1001 (slot 2)
Barcode = 000400
drive 6 (addr 505) access = 1 Contains Cartridge = no
READ_ELEMENT_STATUS complete